Non-stationarity of Climate Extremes: Natural Variability and Anthropogenic Forcings

Climate extremes are the most tangible of societal and economic impacts of climate, warranting a better understanding and quantification of their changing behavior. We analyze the trends and tele-connections of precipitation extremes in a small ensemble (4 realizations) of high-resolution simulations (from the Accelerated Climate Modeling Project [ACME]) and a large ensemble (40 realizations) of low-resolution simulations (from the Community Earth System Model Large Ensemble Project [LENS]). A regionalization framework is applied to improve the sample size of extreme events by using data from neighboring regions with a homogeneous climate. The generalized extreme value theory is used to quantify the statistics and non-stationary behavior of precipitation extremes. The LENS simulations reveal that the trend in the twentieth century is significant in several regions of the world, but the trends are not robust across the simulation ensemble. However, the twenty-first century projections in those simulations forced with RCP 8.5 emissions exhibit a robust increasing trend across the ensemble, suggesting that while natural variability dictated variability of precipitation extremes in the past century, anthropogenic forcing would most likely dominate the variability in the future. Single forcing runs with excessive black carbon aerosol concentrations - replicating nuclear winter scenarios - also suggest a role for aerosols in modulating climate extremes. The computational expense of high-resolution model configurations currently limits the length of the simulation period. ACME baseline high-resolution runs were integrated for the period 1979-2005. We find that such a short duration is not enough to capture trends in precipitation extremes in the observations and thus short high-resolution simulations cannot be properly validated for such trends. However, the high-resolution models are able to improve the simulation of tele-connections of extremes with low-frequency phenomenon like the North Atlantic Oscillation (NAO) as compared to their low-resolution counterparts.